An active power distribution network cluster distributed regulation method based on ADMM
By adopting an active distribution network cluster distributed control method based on ADMM, the problem of large-scale communication and computing pressure under centralized dispatch is solved. It realizes efficient coordination and optimization of distributed resources, reduces computing and communication burden, protects user privacy, and ensures the economy and security of the control scheme.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- STATE GRID ANHUI ELECTRIC POWER CO LTD ANQING POWER SUPPLY COMPANY
- Filing Date
- 2026-03-02
- Publication Date
- 2026-05-29
Smart Images
Figure CN122118906A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of distribution network optimization and intelligent dispatching, and in particular to an active distribution network cluster distributed control method based on ADMM. Background Technology
[0002] With the large-scale integration of distributed resources such as distributed power sources, energy storage devices, and electric vehicles, the operation mode and control model of the distribution network have undergone significant changes. On the one hand, distributed resources are intermittent, random, and uncertain, and their large-scale integration has a significant impact on the safe and stable operation of the distribution network, power quality, and power flow distribution. On the other hand, improving the absorption level of distributed energy and achieving its flexible and efficient operation has become an important requirement for the development of active distribution networks and smart grids.
[0003] Existing power distribution network control systems mostly employ centralized dispatching and management, requiring the uploading of operational data from a large amount of distributed resources to the main station for unified optimization and decision-making. However, due to the diverse types of distributed resources, the large number of connected systems, and the relatively small capacity of individual units, the centralized control model faces challenges such as massive communication data volumes and rapid expansion of computing power, making it difficult to meet real-time and scalability requirements. Furthermore, the centralized uploading of large amounts of user-side operational data also presents challenges such as difficulty in protecting privacy and high information security risks.
[0004] Therefore, there is an urgent need for a control method that can reduce communication and computing pressure while achieving coordinated and optimized operation of distributed resources, so as to improve the distribution network's ability to manage distributed resources and support the development of future active distribution networks and smart grids. Summary of the Invention
[0005] To address the existing problems, this invention provides an active distribution network cluster distributed control method based on ADMM, the specific scheme of which is as follows:
[0006] An active distribution network cluster distributed control method based on ADMM includes the following steps:
[0007] S1. Construct an optimization problem model for the coordinated operation of active and reactive power in an active distribution network. This model includes the objective function and operational constraints.
[0008] S2, introduce virtual nodes at the electrical coupling boundary between clusters and define global coupling variables associated with the virtual nodes; based on the consistency theory, decouple the centralized optimization problem model into local optimization subproblems corresponding to each cluster. The solution of each local optimization subproblem depends only on the internal operation information of the cluster and the boundary constraints determined by the global coupling variables.
[0009] S3. Establish an inter-cluster interaction and coordination mechanism based on the Alternating Direction Multiplier Method (ADMM).
[0010] Preferably, the model in step S1 takes minimizing the total system operating cost as its objective function, which can be expressed as: ;in, This indicates the cost of purchasing electricity from the upstream power grid; This represents the network loss cost of the system; This indicates the cost of energy storage losses; This represents the cost of curtailing electricity; operational constraints include power flow constraints, safe operation constraints, power balance constraints, and distributed resource operation constraints.
[0011] Preferably, the distributed resources include adjustable distributed photovoltaics, distributed energy storage devices, and static var compensators; the operating constraints of the distributed resources include photovoltaic output and reactive power regulation capability constraints, energy storage charging and discharging power and state of charge constraints, and reactive power output constraints of the static var compensators.
[0012] Preferably, in step S2, the global coupling variables are the active power and voltage amplitude at the virtual node of the coupling boundary; the boundary constraint is expressed as follows: for each cluster, the active power and voltage amplitude at the connection between its internal network and the virtual node must be equal to the corresponding values in the global coupling variables.
[0013] Preferably, the mechanism in step S3 includes the following iterative process:
[0014] S31, each cluster solves its corresponding local optimization subproblem in parallel based on the values of the global coupling variables and Lagrange multipliers in the current iteration round, and obtains the optimization decision results and locally calculated values of boundary variables for this cluster.
[0015] S32, based on the local values of the boundary variables calculated by all clusters in step S31, update the global coupling variables using consistency rules;
[0016] S33, Update the Lagrange multipliers based on the difference between the updated global coupling variables and the local values of the boundary variables;
[0017] In step S34, determine whether the original residuals formed by the local values of boundary variables and global coupling variables, and the dual residuals formed by the transformation of Lagrange multipliers, satisfy the convergence condition. If they do, terminate the iteration and output the optimized scheduling scheme for each cluster. Otherwise, feed back the updated global coupling variables and Lagrange multipliers to step S31 and start a new round of iteration.
[0018] Preferably, in step S32, the consistency rule is to take the average value of the local calculations of all relevant clusters for the same global coupling variable.
[0019] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed, performs the method described in any of the preceding claims.
[0020] A computer system includes a processor, a storage medium storing a computer program, and the processor reading from the storage medium and running the computer program to perform the method described in any of the preceding claims.
[0021] The beneficial effects of this invention are as follows:
[0022] 1. Distributed architecture: By partitioning and problem decomposition, large-scale centralized computing tasks are distributed to multiple local agents for parallel processing, which significantly reduces the computing load and solution dimensionality of the main station and improves computing efficiency.
[0023] 2. Limited communication: Agents only need to exchange a small amount of information such as boundary electrical quantities and coordination variables in each iteration. They do not need to disclose detailed internal operating data, which greatly reduces the pressure on the communication network and helps protect user privacy and trade secrets.
[0024] 3. Global optimality guarantee: Under the premise that the problem is convex, the coordination mechanism based on ADMM can guarantee that the distributed iterative results converge to the global optimal solution of the original global optimization problem, thus ensuring the economy and security of the control scheme.
[0025] 4. Plug and play and robustness: New regions or distributed resources can be easily connected to the corresponding local agent without affecting the overall framework. A single region or communication failure does not affect the independent operation of other regions, demonstrating strong system robustness. Attached Figure Description
[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0027] Figure 1 A schematic diagram of the distributed cluster partitioning;
[0028] Figure 2 This is a diagram illustrating cluster decoupling.
[0029] Figure 3 This is a schematic diagram of the IEEE-33 node power distribution system in the embodiment;
[0030] Figure 4 This is a schematic diagram illustrating the prediction of maximum photovoltaic output in the embodiment;
[0031] Figure 5This is a schematic diagram of the distribution network zoning results in the embodiment;
[0032] Figure 6 This is a schematic diagram illustrating the convergence of the algorithm in the embodiment;
[0033] Figure 7 This is a schematic diagram illustrating the power output of each energy storage producer and seller in the embodiment;
[0034] Figure 8 This is a schematic diagram of the node voltage levels in the embodiment. Detailed Implementation
[0035] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0036] This invention proposes an optimization and control method based on local distributed resource information. The operational information of distributed resources is first uploaded to the corresponding edge computing unit, which then aggregates and analyzes the distributed resources within its jurisdiction and performs optimization calculations in conjunction with coordination information from adjacent clusters. This invention employs a consistency-based alternating direction multiplier method to achieve collaborative optimization between different clusters, ensuring independent decision-making for each cluster while achieving system-level coordination. Through this method, distributed resources within each cluster can complete optimization decisions locally, reducing dependence on the master station. Clusters achieve globally optimal or near-optimal distributed scheduling results through limited information exchange.
[0037] An active distribution network cluster distributed control method based on ADMM includes the following steps:
[0038] S1. Construct an optimization problem model for the coordinated operation of active and reactive power in an active distribution network. This model includes an objective function and operational constraints.
[0039] Specifically, due to the low voltage level and high line impedance, the optimized operation of distribution networks often prioritizes economic efficiency while ensuring voltage and power flow safety. The distributed resources considered in this invention include: adjustable distributed photovoltaics, distributed energy storage, and SVC (Supply Dynamic Valve Chassis). The objective function for system operation can be expressed as:
[0040] (1)
[0041] in, This indicates the cost of purchasing electricity from the upstream power grid; This represents the network loss cost of the system; This indicates the cost of energy storage losses; This represents the cost of power curtailment. See (2) for the specific calculation methods for each cost.
[0042] (2)
[0043] Where T represents the scheduling period, and t represents a specific scheduling time segment within the scheduling period. Represents a set of nodes. Represents a set of routes. This represents the price at which electricity is purchased from the upper-level power grid during time period t. This represents the amount of electricity purchased by user i from the upper-level power grid during time period t; Indicates the price of network loss; This indicates the magnitude of the current in line ij during time period t. Indicates the resistance of line ij; This indicates the price per unit of energy storage or release loss. and This represents the energy storage discharge and charging efficiency at node i; and Let represent the discharge and charging power of the energy stored at node i during time period t, respectively; Indicates the unit price of the penalty for abandoning light. This represents the amount of solar power wasted at node i during time period t.
[0044] The constraints include:
[0045] System power flow constraints
[0046] (3)
[0047] in, and They represent Time Node Line The active and reactive power flowing upstream, and They represent the lines respectively. Resistance and reactance, and Representing inflow and outflow nodes respectively. The set, and They represent Time Node The injected active and reactive power; express Time Node The square of the voltage; express Timetable The square of the current flowing through it.
[0048] Safe operation constraints
[0049] To ensure the stable operation of the distribution network, voltage and current must be kept within safe ranges:
[0050] (4)
[0051] in, , and Representing nodes respectively and lines Safety limits for voltage and current.
[0052] Power balance constraints
[0053] (5)
[0054] Distributed photovoltaic power output constraints
[0055] (6)
[0056] in, , and They represent Time period nodes Actual output, theoretical maximum output, and curtailment of distributed photovoltaic power; and This indicates the reactive power output and installed capacity of the photovoltaic system.
[0057] According to the grid connection criteria for distributed photovoltaic (PV) systems, the reactive power regulation capability of inverters for Class B distributed PV systems is as follows:
[0058] (7)
[0059] Distributed energy storage constraints
[0060] (8)
[0061] in, and They represent Time Node The charging and discharging power of energy storage, and These represent its maximum charging and discharging power, respectively; and These represent their charging and discharging states, and they cannot be in both charging and discharging states simultaneously during the same scheduling period. and These represent the charging and discharging efficiencies, respectively. Represents a node Energy storage The state of charge at any given time must not exceed its limit. and And at the beginning of a scheduling cycle and the end of the period The state of charge is consistent, thus maintaining the continuity of the scheduling cycle.
[0062] SVC constraint
[0063] (9)
[0064] in, This represents the reactive power output of node i at time t, which is limited by the installed capacity. between.
[0065] Therefore, the centralized optimization problem of the distribution network can be expressed as finding the minimum value of the objective function (1) under the constraints (2)-(9). It is necessary to collect all the data of the power grid and distributed resources into the main station, process them centrally and distribute them, resulting in a large amount of data transmission and a high model dimension.
[0066] The method proposed in this invention transmits data into, for example Figure 1 The clusters shown are characterized by data aggregation and processing by an edge computing unit in each cluster, with adjacent cluster computing units exchanging a small amount of electrical quantity information at the cluster level.
[0067] S2, introduce virtual nodes at the electrical coupling boundaries between clusters and define global coupling variables associated with the virtual nodes; based on consistency theory, decouple the centralized optimization problem model into local optimization subproblems corresponding to each cluster. The solution of each local optimization subproblem depends only on the internal operating information of the cluster and the boundary constraints determined by the global coupling variables.
[0068] Specifically, the objective mentioned in equation (1) can be decomposed into the sum of sub-objectives of each cluster, i.e. (10):
[0069] (10)
[0070] in, The set represents the cluster. For device-level constraints (5)-(7), it is ensured that each device within each cluster satisfies the condition. For system-level constraints, including power flow constraints, power balance constraints, and safe operation constraints, there is tight coupling between the various clusters, making it impossible to solve based on a single cluster. For example, Figure 1In this context, cluster 1 and cluster 2 share branch 6-7, the current flowing into cluster 2 is the same as the current flowing out of cluster 1, the voltage of node 7 is closely related to the voltage of node 6, etc. To perform calculations for each cluster separately, it is necessary to first decouple each cluster.
[0071] Adopting such Figure 2 The method shown decouples the system by introducing a virtual node k located at the coupled branch. For each cluster, active power and voltage constraints (11) are satisfied. Due to Ohm's law, once active power and voltage are constrained, current and reactive power are naturally constrained, so there is no need to introduce other variables. Let the power and voltage at this point be global variables. .
[0072] (11)
[0073] Therefore, for each cluster, given the global variables, the problem to be solved changes from (1) to (12):
[0074] (12)
[0075] The constraints include all system-level constraints (3)-(5) and device-level constraints (6)-(9), as well as dual constraints (11). In this problem, since the global variables coupled with other clusters are known, only the information within the cluster is needed to solve the problem, which aligns with the goal of local decision-making. However, there is no accurate basis for determining the global coupling variables, and it is impossible to guarantee the optimal collaboration among the various clusters. It is also necessary to exchange information among the clusters in S3 to optimize the global variables and achieve global optimum.
[0076] S3 utilizes information exchange and coordination between clusters to solve the various sub-problems established in S2, that is, to establish an inter-cluster interaction and coordination mechanism based on the Alternating Direction Multiplier Method (ADMM).
[0077] Specifically, to ensure optimal collaboration among clusters in S2, this invention uses the ADMM method to guarantee convergence of each cluster towards the optimal state. Lagrange variables are introduced. and penalty coefficient Then the Lagrange function of (10) can be expressed as:
[0078] (13)
[0079] Since each cluster has constraints (11), when the optimal solution is reached... It is also 0. The problem that each cluster needs to solve becomes:
[0080] (14)
[0081] The specific process using the ADMM method is as follows:
[0082] 1. Initialize all global variables, Lagrange coefficients, and penalty coefficients, denoted as... and , Let be the iteration number, which is 0 at this point. Based on these known quantities, the optimal scheduling scheme and variables of each cluster in problem (14) can be solved. Let the power and voltage at the coupled node be denoted as . and ;
[0083] 2. Since each cluster in (14) is a convex optimization problem and the problems of each cluster are symmetrical, the global variables can be solved using KKT conditions:
[0084] (15)
[0085] Update the Lagrange coefficients according to the ADMM calculation process:
[0086] (16)
[0087] 3. After each iteration, calculate the original residual and the dual residual. Original residual:
[0088] (17)
[0089] 4. When both the dual residual and the original residual are less than the threshold, the iteration ends, and the scheduling result obtained for each region is the global optimal solution. If the two residuals do not reach the convergence threshold, the iteration continues based on the updated... Go back to step one and recalculate.
[0090] The following is an example:
[0091] Adopting such Figure 3 The IEEE-33 node power distribution system shown is illustrated, with line impedance and load information calculated according to the IEEE-33 standard. The total installed capacity of distributed photovoltaic and energy storage is 2400 kW and 800 kW, respectively. Specific installed capacity data is shown in Table 1, and price data is shown in Table 2.
[0092] Table 1 Specific Installation Data
[0093]
[0094] Table 2 Price-related parameters
[0095]
[0096] The predicted maximum output curves for distributed photovoltaic power at each node are as follows: Figure 4 As shown.
[0097] Systems can be categorized based on electrical distance as follows: Figure 5 The seven regions shown are connected to the upstream power grid, with node 0 being the node that connects to the upstream power grid.
[0098] Algorithm convergence status, such as Figure 6 Regarding distributed solution efficiency, clustered coupled electrical quantities converge faster; the residual can reach a certain level after 15 iterations. Left and right, all elements in the global variable set , The sums are 190.48 and 4.83 respectively, when the residuals reach After the level, the relative error is approximately This can be considered negligible.
[0099] This invention can effectively reduce the communication burden and computational complexity in the distribution network control process, improve the security and economy of distributed resource operation, and enhance the privacy protection of user data, providing an efficient and scalable technical support for the optimized operation of distributed resources in the distribution network.
[0100] The present invention also discloses a computer-readable storage medium and a computer system. The medium stores a computer program, which, upon execution, performs the method described in any of the preceding claims. A computer system includes a processor and a storage medium, the storage medium storing a computer program, and the processor reading from and running the computer program to perform the method described in any of the preceding claims.
[0101] Those skilled in the art will further appreciate that the various illustrative logic blocks, modules, circuits, and algorithm steps described in conjunction with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or a combination of both. To clearly illustrate this interchangeability between hardware and software, the various illustrative components, blocks, modules, circuits, and steps are described above in a generalized manner in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system. Those skilled in the art may implement the described functionality in different ways for each specific application, but such implementation decisions should not be construed as departing from the scope of the invention.
[0102] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A distributed control method for active distribution network clusters based on ADMM, characterized in that, Includes the following steps: S1. Construct an optimization problem model for the coordinated operation of active and reactive power in an active distribution network. This model includes the objective function and operational constraints. S2, introduce virtual nodes at the electrical coupling boundary between clusters and define global coupling variables associated with the virtual nodes; based on the consistency theory, decouple the centralized optimization problem model into local optimization subproblems corresponding to each cluster. The solution of each local optimization subproblem depends only on the internal operation information of the cluster and the boundary constraints determined by the global coupling variables. S3. Establish an inter-cluster interaction and coordination mechanism based on the Alternating Direction Multiplier Method (ADMM). After each cluster solves its subproblems in parallel, it alternately updates the global coupling variables and Lagrange multipliers. Through limited information interaction and iteration, it drives the local decisions of all clusters to converge to the global optimal solution of the system.
2. The method according to claim 1, characterized in that: The model described in step S1 takes minimizing the total operating cost of the system as its objective function, which can be expressed as: ;in, This indicates the cost of purchasing electricity from the upstream power grid; This represents the network loss cost of the system; This indicates the cost of energy storage losses; This represents the cost of curtailing electricity; operational constraints include power flow constraints, safe operation constraints, power balance constraints, and distributed resource operation constraints.
3. The method according to claim 2, characterized in that: The distributed resources include adjustable distributed photovoltaics, distributed energy storage devices, and static var compensators; the operating constraints of the distributed resources include photovoltaic output and reactive power regulation capability constraints, energy storage charging and discharging power and state of charge constraints, and reactive power output constraints of static var compensators.
4. The method according to claim 1, characterized in that: In step S2, the global coupling variables are the active power and voltage amplitude at the virtual node of the coupling boundary; the boundary constraint is expressed as follows: for each cluster, the active power and voltage amplitude at the connection between its internal network and the virtual node must be equal to the corresponding values in the global coupling variables.
5. The method according to claim 1, characterized in that: The mechanism described in step S3 includes the following iterative process: S31, each cluster solves its corresponding local optimization subproblem in parallel based on the values of the global coupling variables and Lagrange multipliers in the current iteration round, and obtains the optimization decision results and locally calculated values of boundary variables for this cluster. S32, based on the local values of the boundary variables calculated by all clusters in step S31, update the global coupling variables using consistency rules; S33, Update the Lagrange multipliers based on the difference between the updated global coupling variables and the local values of the boundary variables; In step S34, determine whether the original residuals formed by the local values of boundary variables and global coupling variables, and the dual residuals formed by the transformation of Lagrange multipliers, satisfy the convergence condition. If they do, terminate the iteration and output the optimized scheduling scheme for each cluster. Otherwise, feed back the updated global coupling variables and Lagrange multipliers to step S31 and start a new round of iteration.
6. The method according to claim 1, characterized in that: In step S32, the consistency rule is to take the average value of the local calculation of the same global coupling variable by all related clusters.
7. A computer-readable storage medium, characterized in that: The medium contains a computer program, which, when run, performs the method as described in any one of claims 1 to 6.
8. A computer system, characterized in that: It includes a processor and a storage medium, on which a computer program is stored, and the processor reads from the storage medium and runs the computer program to perform the method as described in any one of claims 1 to 6.